Keywords

1 Introduction

Today for meeting the customer’s expectations, the availability of machines is very much essential. When a gas turbine goes down, it would result in power shut down. This would cause major inconvenience for the customers. If a gas turbine becomes unavailable, then there will be downtime and associated waiting time for its customers. That is, the customer will have to wait for till the gas turbine becomes operational. Amount of waiting time is estimated by the severity of the problem faced by the gas turbine. In this scenario, there will be idle time of the gas turbine and the operator as well. For example, in a manufacturing company, when a gas turbine comes non-functional, it may impact other machines down the manufacturing line. Thus, when a gas turbine becomes non-operational, it would result in affecting the productivity of the manufacturing line. This would result in customer dissatisfaction, and companies will lose its reputation. This clearly shows the significance of properly maintaining a gas turbine. Thus, the productivity of a manufacturing line depends on the availability of gas turbines. For increasing the turbine’s availability, selection of a right maintenance policy is very much essential. Selection of right maintenance policy will bring down the maintenance cost of the gas turbine. Additionally, selecting right type of maintenance for a given machine will bring down the workplace-related accidents, and this will enhance employee well-being and comfort level. When a machine becomes available, it would help a machine tool operator in meeting his production targets. Thus, it will make the operator happy and enhance his productivity.

Maintenance can be classified based on time and (Ben et al. 2016; Shayesteh et al. 2018). There is a huge need for reducing the maintenance cost of machines in general and gas turbine in particular. Not much work has been reported in literature, about the economical maintenance of a gas turbine. This has motivated the author in focusing on maintenance of gas turbine. The research findings provide guidelines for selecting maintenance policies for gas turbines.

2 Literature Survey

Aerospace companies have had problems with the conventional maintenance methods, and this has led to the research the emergence of RCM in 1950.

Barlow et al. (1975) and Birnbaum (1969) proposed a technique of identifying components which are significant to a machine. Significant or critical components, when fail, become impossible to meet customer’s expectations (Jeyamala et al. 2013). Saaty (1996) has designed a new method called analytic network process (ANP). This method was useful in identifying the critical components. ANP has been employed by many researchers as a decision-making tool (Dorri 2014; Sadeghi and Manesh 2015; Tajadod et al. 2011). A new method was proposed by Moslemi et al. (2017). This technique was used to determine the preventive maintenance schedule for an equipment.

Hamzeh et al. (2015) and Marton et al. (2016) have explored techniques of addressing the maintenance-related problems of an aging distribution system. They have studied in detail of the aging process and its effects on equipment maintenance.

Carnero et al. (2017) have explored electric power systems and proposed right methods for addressing their maintenance problems. Silvestri et al. (2012) have explored a manufacturing industry and designed and developed a maintenance framework for enhancing worker’s safety. Wang et al. (2001) have proposed a new technique for evaluating the reliability of a CNC machine in economical manner. Carot et al. (2000) in their study had used non-repairable parts and performed sensitivity analysis for assessing the relative prominence of different parts of a complex network. To overcome the drawbacks of RCM (Alrifaey et al. 2019; Zhou et al. 2016; Liu et al. 2019; Carmignani 2009; Crocker and Kumar 2000; Campbell et al. 2016), a novel method is proposed for selecting the maintenance policy for a gas turbine. Thus, it is evident that RCM technique was not explored for the maintenance of gas turbines. In this context, the current research work becomes very significant. The research outputs are very much useful for both academicians and practitioners in pursuing further research.

3 Methodology

The current research work uses the steps described in the following paragraphs.

3.1 System Selection

In the current research, a gas turbine used in oil and gas processing plant is considered as a case study (power generation 600 MW). Gas turbine (Fig. 1) assembly consists of compressor, combustion system, power turbine, generator, and integral equipment (e.g., fuel, oil, and electrical subsystems) are considered as a system.

Fig. 1
figure 1

Gas turbine-based electric power generation

Full size image

3.2 Analytical Network Process (ANP)

When multiple alternatives exist in a decision, weights can be assigned to each of the alternatives using analytical network process (ANP). That is, ANP would help in classification. Thus, ANP would help in recognizing the critical components. The method was designed by Saaty (1996) in the 1970s. Analytical network process has to follow principles—given problem is set-up in a hierarchical manner, and by paired comparison, relative weights are determined, logical consistency through measurements. MATLAB software is used for making calculations related to ANP process. The technique helps in calculating the weights for different alternatives. The technique makes use of expert’s experience in quantifying weights of various decisions through pairwise comparisons. After conducting interviews with experts and based on their feedback, the following criteria clusters are identified: (1) cost: A component is considered critical if it has higher maintenance cost in comparison to other components, (2) complexity: An assembly with large number of components with high failure rates is considered more critical, (3) maintainability: An assembly with large downtime is considered more critical than others, (4) safety: It is important to consider the safety aspect while identifying components which are critical to the functioning of a module, and (5) reliability: It shows the extent to which the equipment is dependable. A more dependable equipment will ensure meeting of the customer expectation in a safe and sustainable way.

ANP consists of the following steps.

Step 1: to calculate the relative importance of each of the five elements—cost, complexity, maintainability, reliability, and safety.

Step 2: computing the relative importance of the components specified.

Step 3: construction of super-matrix that is constructed based on factors and its elements derived from the pairwise comparison.

Step 4: in this step, super-matrix computed in the previous step is multiplied by the priority of the factors.

Step 5: to calculate the inconsistency of the pairwise comparison matrix

$$Compatibility\,Index(CI)=\frac{\lambda max-n}{\mathrm{n}-1}$$
(1)

In Eq. (1), n represents the matrix aspect; the numerator shows the extent of incompatibility existing in the matrix (Birnbaum 1969).

Ratio of compatibility is calculated by using Eq. (2)

$$CR=\frac{CI}{\mathrm{RI}}$$
(2)

If CR <  = 00.1, matrix is said to be compatible.

Step 6: to calculate the weight of the super-matrix by using the following equation.

$${\text{Aw}} = \lambda_{\text{max} } {\text{W}}$$
(3)

where A is the matrix of the paired-comparison, λmax is the maximum eigen value, and W is the eigen vector.

Step 7: normalize the matrix.

Step 8: to calculate the weights of final limit matrix (W).

$${\text{WL}} = {\text{Wr}}^{2{\text{k}} + 1}$$
(4)

4 Results and Discussions

The objective of performing the current research is to figure out right maintenance policy for different components of gas turbine used in an electricity generation in oil and gas company. The main problem here is that there are multiple components in a gas turbine, and the company has a maintenance budget. It was not possible to maintain all the components of a gas turbine company within the allocated budget. The company was looking for ways to improve service by maintenance without exceeding maintenance budget. In this context, the present research work assumes special significance.

Current research work uses analytical network process, and steps are described in previous section (Sect. 3, of this paper). For performing the analytical network process, MATLAB software is used. The results of the ANP are given in Tables 1, 2, and 3.

Table 1 Pairwise criteria matrix
Full size table
Table 2 ANP super-matrix (criteria/alternatives)
Full size table
Table 3 Weights—super-matrix (W)
Full size table

Table 1 shows the criteria-wise pairwise comparison to goal. The table shows the relative importance of the different criteria with respect to each other. In Table 1, G1–G5 are the different criteria identified by experts (G1: cost; G2: maintainability; G3: complexity; G4: reliability; G5: safety). Experts selected are seniors working in different departments of the oil and gas company. Five experts are chosen for the current research work. Information is obtained from experts by conducting interviews.

Inconsistency ratio was determined, and compatibility was ensured. Similarly, pairwise criteria matrix steps give the other columns of the super-matrix (Eq. 3). Table 2 shows the super-matrix for criteria and alternatives. This is obtained by using Eq. (3). All these calculations were done using MATLAB software. After prioritization of criteria and alternatives, decision tree was implemented to select the maintenance policy for the different failure modes. Table 4 shows the failure modes, its ranking, and the maintenance policy. Failure modes with top ranking will attract condition-based maintenance. As, these failure modes would result in severe ill-effects.

Table 4 Selection of maintenance policies
Full size table

5 Conclusion

Maintenance (conditional maintenance/scheduled preventive maintenance/preventive maintenance) helps in extending the useful life of the machine tools in a manufacturing company. In the present research work, a gas turbine-based electricity generator in an oil and gas company is considered as a case study. An effort is made to apply the analytical hierarchy process for selecting the appropriate policy for maintaining the different components of a gas turbine.

Enhancing the availability will ensure meeting customer orders. A gas turbine consisted of different components. In the oil and gas company (case study), budget of maintenance was limited. Thus, it was not possible to focus on all the components of gas turbine for enhancing the availability of the gas turbine. In this context, critical components had to be identified for reducing the maintenance cost.

Analytical network process (ANP) makes it possible to assign weights to multiple alternatives. Thus, it helps in selecting or classifying alternatives. The technique helps in calculating the weights for different decisions. The technique makes use of expert’s experience in quantifying weights of various decisions through pairwise comparisons. After conducting interviews with experts and based on their feedback, the following criteria clusters are identified: (1) cost cluster: A component is considered critical if it has higher maintenance cost in comparison to other components; (2) complexity cluster: A machine with large number of components with high failure rates is considered more critical; (3) maintainability cluster: A machine with large downtime is considered more critical than others; and (4) safety cluster: It is important to consider the safety aspect while identifying components which are critical to the functioning of a module; (5) reliability: It measures the extent of dependability of equipment. Having a more dependable equipment would ensure meeting customer expectations in safe and sustainable way.

Pairwise comparisons have showed the relative importance of each component with respect other components. Overall importance and the criticality analysis showed that certain components—fire detector, vibration and displacement transmitter, purging controller, bearing and its transmitter, and temperature transmitter—are critical for the effective functioning of the gas turbine. As these components are critical and since the availability of the gas turbine depends on these components, condition-based maintenance policy is preferred. Certain components, e.g., transmitter and control valve had scored less criticality rating. Thus, transmitter and control valve are not considered as a critical component of the gas turbine, and preventive maintenance is preferred. For certain components, control system and control panel, because they are not critical, corrective maintenance (CM) policy is preferred. The current research work showed that not all components require same type of maintenance policies. Thus, by focusing only on critical components of gas turbine, there will be a reduction of total maintenance cost.